sunholo 0.80.6__tar.gz → 0.81.1__tar.gz

{sunholo-0.80.6 → sunholo-0.81.1}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.80.6
+Version: 0.81.1
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Home-page: https://github.com/sunholo-data/sunholo-py
-Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.80.6.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.81.1.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

sunholo-0.81.1/setup.cfg

@@ -0,0 +1,16 @@
+[metadata]
+description-file = README.md
+
+[mypy]
+files = sunholo, tests
+python_version = 3.9
+disallow_untyped_calls = True
+warn_return_any = True
+warn_unused_ignores = True
+strict_optional = True
+ignore_missing_imports = True
+
+[egg_info]
+tag_build = 
+tag_date = 0
+

{sunholo-0.80.6 → sunholo-0.81.1}/setup.py

@@ -1,7 +1,7 @@
 from setuptools import setup, find_packages
 
 # Define your base version
-version = '0.80.6'
+version = '0.81.1'
 
 setup(
     name='sunholo',

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/auth/refresh.py

@@ -4,8 +4,6 @@ import os
 import google.auth 
 from google.auth.transport import requests
 from ..utils.gcp import is_running_on_gcp
-
-
 from ..custom_logging import log
 
 def get_default_email():
@@ -26,7 +24,7 @@ def get_default_email():
             return None
     
     log.info(f"Found default email: {service_account_email=} for {project_id=}")
-    return service_account_email
+    return service_account_email, gcs_credentials
 
 def get_default_creds():
     gcs_credentials = None

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/components/retriever.py

@@ -170,4 +170,5 @@ def process_retrieval(retriever_list: list, config: ConfigManager):
         base_compressor=pipeline, base_retriever=lotr, 
         k=k_override)
     
+    log.info(f"Returning Langchain retrieval object: {retriever}")
     return retriever

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/database/__init__.py

@@ -4,3 +4,4 @@ from .database import setup_database
 from .database import return_sources_last24
 from .database import delete_row_from_source
 from .static_dbs import get_db_directory
+from .alloydb_client import AlloyDBClient

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/database/alloydb.py

@@ -14,6 +14,8 @@ from .alloydb_client import AlloyDBClient
 from ..custom_logging import log
 from ..utils.config import load_config_key
 
+from typing import List
+
 
 def create_alloydb_engine(vector_name):
 
@@ -231,8 +233,11 @@ async def load_alloydb_sql_async(sql, vector_name):
     
     return documents
 
-def and_or_ilike(sources, search_type="OR", operator="ILIKE"):
-    unique_sources = set(sources.split())
+def and_or_ilike(sources:List[str], search_type:str="OR", operator:str="ILIKE"):
+    if not isinstance(sources, list) or not all(isinstance(source, str) for source in sources):
+        raise TypeError("The `sources` argument must be a list of strings.")
+    
+    unique_sources = set(sources)
     # Choose the delimiter based on the search_type argument
     delimiter = ' AND ' if search_type.upper() == "AND" else ' OR '
 

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/database/alloydb_client.py

@@ -1,3 +1,4 @@
+import os
 try:
     import pg8000
     import sqlalchemy
@@ -9,6 +10,7 @@ except ImportError:
 
 from .database import get_vector_size
 from ..custom_logging import log
+from ..utils import ConfigManager
 
 class AlloyDBClient:
     """
@@ -35,11 +37,12 @@ class AlloyDBClient:
     """
 
     def __init__(self, 
-                 project_id: str, 
-                 region: str, 
-                 cluster_name:str, 
-                 instance_name:str, 
-                 user:str, 
+                 config:ConfigManager,
+                 project_id: str=None, 
+                 region: str=None, 
+                 cluster_name:str=None, 
+                 instance_name:str=None, 
+                 user:str=None, 
                  password=None, 
                  db="postgres"):
         """Initializes the AlloyDB client.
@@ -51,6 +54,23 @@ class AlloyDBClient:
          - password (str): The database user's password.
          - db_name (str): The name of the database.
         """
+        if config is None:
+            if project_id is None or region is None or cluster_name is None or instance_name is None:
+                raise ValueError("Must specify config or project_id, region, cluster_name, instance_name")
+        if config:
+            alloydb_config = config.vacConfig("alloydb_config")
+            if not alloydb_config:
+                raise ValueError("Must specify vac.alloydb_config")
+            project_id = alloydb_config["project_id"]
+            region = alloydb_config["region"]
+            cluster_name = alloydb_config["cluster"]
+            instance_name = alloydb_config["instance"]
+
+        ALLOYDB_DB = os.environ.get("ALLOYDB_DB")
+        if ALLOYDB_DB is None and alloydb_config.get("database") is None:
+            log.warning("Could not locate ALLOYDB_DB environment variable or 'alloydb_config.database'")
+        
+        self.database = alloydb_config.get("database") or ALLOYDB_DB,
         self.connector = Connector()
         self.inst_uri = self._build_instance_uri(project_id, region, cluster_name, instance_name)
         self.engine = self._create_engine(self.inst_uri, user, password, db)
@@ -100,6 +120,80 @@ class AlloyDBClient:
                 conn.close()
 
         return result
+    
+    async def execute_sql_async(self, sql_statement):
+        """Executes a given SQL statement asynchronously with error handling."""
+        sql_ = sqlalchemy.text(sql_statement)
+        result = None
+        async with self.engine.connect() as conn:
+            try:
+                log.info(f"Executing SQL statement asynchronously: {sql_}")
+                result = await conn.execute(sql_)
+            except DatabaseError as e:
+                if "already exists" in str(e):
+                    log.warning(f"Error ignored: {str(e)}. Assuming object already exists.")
+                else:
+                    raise
+            finally:
+                await conn.close()
+
+        return result
+
+    async def get_sources_from_docstore_async(self, sources, vector_name, search_type="OR", just_source_name=False):
+        """Fetches sources from the docstore asynchronously."""
+        if just_source_name:
+            query = self._list_sources_from_docstore(sources, vector_name=vector_name, search_type=search_type)
+        else:
+            query = self._get_sources_from_docstore(sources, vector_name=vector_name, search_type=search_type)
+
+        if not query:
+            return []
+
+        documents = await self.execute_sql_async(query)
+        return documents
+
+    def _get_sources_from_docstore(self, sources, vector_name, search_type="OR"):
+        """Helper function to build the SQL query for fetching sources."""
+        if not sources:
+            log.warning("No sources found for alloydb fetch")
+            return ""
+
+        table_name = f"{vector_name}_docstore"
+
+        conditions = self._and_or_ilike(sources, search_type=search_type)
+
+        query = f"""
+            SELECT * 
+            FROM {table_name}
+            WHERE {conditions}
+            ORDER BY langchain_metadata->>'objectId' ASC
+            LIMIT 500;
+        """
+
+        return query
+
+    def _list_sources_from_docstore(self, sources, vector_name, search_type="OR"):
+        """Helper function to build the SQL query for listing sources."""
+        table_name = f"{vector_name}_docstore"
+
+        if sources:
+            conditions = self._and_or_ilike(sources, search_type=search_type)
+            query = f"""
+                SELECT DISTINCT langchain_metadata->>'objectId' AS objectId
+                FROM {table_name}
+                WHERE {conditions}
+                ORDER BY langchain_metadata->>'objectId' ASC
+                LIMIT 500;
+            """
+        else:
+            query = f"""
+                SELECT DISTINCT langchain_metadata->>'objectId' AS objectId
+                FROM {table_name}
+                ORDER BY langchain_metadata->>'objectId' ASC
+                LIMIT 500;
+            """
+
+        return query
 
     @staticmethod
     def _and_or_ilike(sources, search_type="OR", operator="ILIKE"):

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/discovery_engine/get_ai_search_chunks.py

@@ -43,7 +43,7 @@ def get_all_chunks(question:str, config:ConfigManager):
     return None
 
 def get_chunks(question, vector_name, num_chunks):
-    de = DiscoveryEngineClient(vector_name, project_id=get_gcp_project())
+    de = DiscoveryEngineClient(vector_name, project_id=get_gcp_project(include_config=True))
     try:
         return de.get_chunks(question, num_previous_chunks=num_chunks, num_next_chunks=num_chunks)
     except Exception as err:

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/gcs/add_file.py

@@ -22,7 +22,7 @@ except ImportError:
     storage = None
 
 from ..custom_logging import log
-from ..utils import load_config_key, ConfigManager
+from ..utils import ConfigManager
 
 
 def handle_base64_image(base64_data: str, vector_name: str, extension: str):

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/gcs/download_url.py

@@ -1,3 +1,9 @@
+from __future__ import annotations
+from typing import Optional, Tuple, TYPE_CHECKING
+if TYPE_CHECKING:
+    from PIL import Image
+    from google.cloud.storage.bucket import Bucket
+
 import os
 from urllib.parse import quote
 from datetime import datetime, timedelta
@@ -6,25 +12,20 @@ from datetime import datetime, timedelta
 from google.auth.exceptions import RefreshError
 
 try:
-    from google.cloud import storage 
+    from google.cloud import storage
 except ImportError:
     storage = None
 
 from ..custom_logging import log
-from ..utils.gcp import is_running_on_gcp
-from ..auth.refresh import refresh_credentials, get_default_creds, get_default_email
+from ..auth.refresh import refresh_credentials, get_default_email
 from io import BytesIO
 try:
     from PIL import Image
 except ImportError:
     Image = None
 
-gcs_credentials = None
-project_id = None
-gcs_client = None
-gcs_bucket_cache = {}
 
-def get_image_from_gcs(gs_uri: str):
+def get_image_from_gcs(gs_uri: str) -> Image.Image: # type: ignore
     """Converts image bytes from GCS to a PIL Image object."""
     image_bytes = get_bytes_from_gcs(gs_uri)
     if not Image:
@@ -35,7 +36,7 @@ def get_image_from_gcs(gs_uri: str):
     except IOError as e:
         raise ValueError("Unable to open image from bytes:", e)
 
-def get_bytes_from_gcs(gs_uri):
+def get_bytes_from_gcs(gs_uri) -> Optional[bytes]:
     """
     Downloads a file from Google Cloud Storage and returns its bytes.
 
@@ -73,22 +74,39 @@ def get_bytes_from_gcs(gs_uri):
         log.error(f"Error downloading file from GCS: {str(err)}")
         return None
 
+gcs_bucket_cache = {}
+def get_bucket(bucket_name: str) -> Optional[Bucket]:
+    """
+    Gets a Cloud Storage bucket and initialised GCS client
+
+    Args:
+        bucket_name: Name of the bucket 
+
+    Returns:
 
-if is_running_on_gcp():
-    # Perform a refresh request to get the access token of the current credentials (Else, it's None)
-    gcs_credentials, project_id = get_default_creds()
-    # Prepare global variables for client reuse
+    """
     if storage:
         gcs_client = storage.Client()
+    else:
+        raise ImportError("google storage pip required - install via `pip install sunholo[gcp]`")
 
-def get_bucket(bucket_name):
     if bucket_name not in gcs_bucket_cache:
         gcs_bucket_cache[bucket_name] = gcs_client.get_bucket(bucket_name)
     return gcs_bucket_cache[bucket_name]
 
-def sign_gcs_url(bucket_name:str, object_name:str, expiry_secs = 86400):
+def sign_gcs_url(bucket_name:str, object_name:str, expiry_secs:int = 86400) -> Optional[str]:
+    """
+    Creates a signed URL so that users can download a file from Google Cloud Storage without authentication
+
+    Args:
+        bucket_name: Name of the bucket where the object lies
+        object_name: Object within the bucket
+        expiry_secs: How long the link will be valid - default 24hrs
     
-    service_account_email = get_default_email()
+    Returns:
+        str: The signed URL or None if not avialable
+    """
+    service_account_email, gcs_credentials = get_default_email()
 
     expires = datetime.now() + timedelta(seconds=expiry_secs)
 
@@ -115,7 +133,7 @@ def sign_gcs_url(bucket_name:str, object_name:str, expiry_secs = 86400):
         return None
 
 
-def construct_download_link(source_uri: str) -> tuple[str, str, bool]:
+def construct_download_link(source_uri: str) -> Tuple[str, str, bool]:
     """Creates a viewable Cloud Storage web browser link from a gs:// URI.""" 
     if not source_uri.startswith("gs://"):
         return source_uri, source_uri, False  # Return the URI as is if it doesn't start with gs://
@@ -132,7 +150,7 @@ def construct_download_link(source_uri: str) -> tuple[str, str, bool]:
     return construct_download_link_simple(bucket_name, object_name)
 
 
-def construct_download_link_simple(bucket_name:str, object_name:str) -> tuple[str, str, bool]:
+def construct_download_link_simple(bucket_name:str, object_name:str) -> Tuple[str, str, bool]:
     """Creates a viewable Cloud Storage web browser link from a gs:// URI.
 
     Args:
@@ -142,12 +160,14 @@ def construct_download_link_simple(bucket_name:str, object_name:str) -> tuple[st
         A URL that directly access the object in the Cloud Storage web browser.
     """
 
+    if object_name.startswith("gs://"):
+        object_name = object_name.replace("gs://", "https://storage.cloud.google.com")
+
     public_url = f"https://storage.cloud.google.com/{bucket_name}/{quote(object_name)}"
     filename = os.path.basename(object_name)
-    signed = False
-    return public_url, filename, signed
+    return public_url, filename, False
 
-def parse_gs_uri(gs_uri: str) -> tuple[str, str]:
+def parse_gs_uri(gs_uri: str) -> Tuple[str, str]:
     """Parses a gs:// URI into the bucket name and object name.
 
     Args:

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/utils/config_class.py

@@ -37,9 +37,10 @@ class ConfigManager:
         self.config_folder = os.getenv("VAC_CONFIG_FOLDER", os.getcwd())
         self.local_config_folder = local_config_folder
         self.configs_by_kind = self.load_all_configs()
+        self.validate = validate
 
         test_agent = self.vacConfig("agent")
-        if not test_agent and self.vector_name != "global" and validate:
+        if not test_agent and self.vector_name != "global" and self.validate:
             print(f"WARNING: No vacConfig.agent found for {self.vector_name} - are you in right folder? {local_config_folder=} {self.config_folder=}")
 
     def load_all_configs(self):
@@ -125,7 +126,6 @@ class ConfigManager:
             else:
                 config = yaml.safe_load(file)
         self.config_cache[filename] = (config, datetime.now())
-        log.debug(f"Loaded and cached {config_file}")
         if is_local:
             log.warning(f"Local configuration override for {filename} via {self.local_config_folder}")
         return config

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo/utils/gcp_project.py

@@ -24,6 +24,7 @@ def get_gcp_project(include_config=False):
         gcp_config = load_config_key("gcp_config", "global", "vacConfig")
         if gcp_config:
             if gcp_config.get('project_id'):
+                logging.info("Using project_id from vacConfig.gcp_config.project_id")
                 return gcp_config.get('project_id')
 
     project_id = get_env_project_id()

{sunholo-0.80.6 → sunholo-0.81.1}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.80.6
+Version: 0.81.1
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Home-page: https://github.com/sunholo-data/sunholo-py
-Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.80.6.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.81.1.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

sunholo-0.80.6/setup.cfg

@@ -1,7 +0,0 @@
-[metadata]
-description-file = README.md
-
-[egg_info]
-tag_build = 
-tag_date = 0
-